It is now commonplace to see maps of human brain activity created using functional magnetic resonance imaging (fMRI). This method has greatly advanced our understanding of how activity in the brain corresponds to particular percepts, actions, and thoughts. More recently, even the spontaneous activity in the brain in the absence of any task or stimulus has cast light on the functional organization of the brains intrinsic networks and operational principles. At its core, fMRI is a readout of local disturbances to an artificially induced magnetic field brought about by local changes in blood flow, which is in turn linked to neural activity. Following the initial discovery of this functional MRI method more than two decades ago, it was not immediately obvious that neural activity and blood flow should so reliably coupled that the latter could act as a reliable surrogate signal for the former. Now this coupling is nearly taken for granted. Yet blood flow and neural activity are clearly different, and this fact has continued to frustrate a generation of neuroscientists who have sought a convenient means of translating fMRI signals into neural terms. Even in theory, this problem seems and is impossible: how can one pinpoint a unique pattern of activity among millions of neurons that maps to a slow series of blood-based changes in image intensity within a volume of the brain? Despite the inherently underdetermined nature of the problem, this is an important area of study simply because any clues can have wide-reaching consequences for interpreting experiments in humans. While our laboratory does not study neurovascular coupling per se, we do undertake experiments that bring new insights into the interpretation of the hemodynamic fMRI signal. Our recent work in this area has focused on two points. First, we are studying the nature of local neural diversity of in the spiking responses to different types of signals, and how this bears on the hemodynamic responses from the same voxel or area. Second, we are investigating the relationship between activity in large-scale functional MRI networks across the brain to local neural activity measured at a single position. In one of the papers published in the last years (Park et al., Neuron 2017), we discovered that neurons from within a single voxel were highly varied in their responses to naturalistic movies. Important from the perspective of the basis of the fMRI signal, we further found the local blood flow activity followed only a small proportion (16%) of 150 neurons tightly packed within a single voxel < 1mm3. The time courses of these special neurons closely matched both the local field potential (LFP) signal and the vascular response, suggesting that they may have a particular importance for local vascular control. This result may help explain why there is often a discrepancy between local spiking activity on the one hand and LFP and fMRI activity on the other, a point we have previously made in the laboratory (Maier et al. Nature Neuroscience 2008). We are currently pursuing this research direction through the use of simultaneous single-unit recordings inside the MR scanner. Over the years, we have developed and acquired many of the necessary components (MR-compatible electrodes and microdrive, suitable RF coils, preamplifiers, cables, and filters) to achieve such simultaneous recording. With a new postdoc in the laboratory, we hope to understand the nature of such neural diversity under multiple conditions, including spontaneous activity and the free viewing of trial-unique natural videos. Comparing single unit and field potential recordings with fMRI time courses under these conditions requires simultaneous data collection. The second area in which we have investigated the basis of the fMRI signal involves the manipulation of the basal forebrain, a small areas that projects broadly to the cerebral cortex. In our first published paper on this, we examined the effects of reversibly inactivating the basal forebrain using pharmacological agents (Turchi, Chang, et al.,2018). Our initial experiments inactivated portions of this structure in animals undergoing fMRI testing. They suggested that the basal forebrain is centrally involved in regulating spontaneous signals throughout the telencephalon, at least during rest. This effect is particularly pronounced during transitions of arousal, gauged by eye opening and closure. Through collaboration (Liu X et al., manuscript under review), we have also found that the spontaneous signals in the human cerebral cortex are likely also driven by basal forebrain input. These studies paint a new picture of spontaneous brain activity that consists of two rather different components: the first global component is orchestrated regionally by the basal forebrain long-range projections, whereas the second local component is more strongly determined by other factors, including but not limited to anatomical projections. At present, we are beginning a new study that will employ viral optogenetic methods in the macaque to perturb different basal forebrain subdivisions in monkeys at rest and during the performance of a range of behavioral tasks. We also plan to record single unit activity from the basal forebrain wirelessly while the monkey is in its cage in order to establish long-term dynamics of activity within this small but influential area.